Learning Robust Facial Landmark Detection via Hierarchical Structured Ensemble

Zou, Xu; Yan, Luxin; Zhao, Xiangyun; Zhou, Jiahuan; Wu, Ying

doi:10.1109/iccv.2019.00023

Cited by 57 publications

(16 citation statements)

References 49 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Recently, the landmark detections with high accuracy and strong robustness are mainly conducted by predicting heatmap with bottom-up and top-down network structure [8]- [10]. A. Bulat and G. Tzimiropoulos explored 3D facial landmark detection with the structure [11].…”

Section: A Cnn-based Methodsmentioning

confidence: 99%

See 1 more Smart Citation

Fast Head Pose Estimation via Rotation-Adaptive Facial Landmark Detection for Video Edge Computation

et al. 2020

View full text Add to dashboard Cite

The human head pose estimation is an important and challenging problem, which provides the estimation of the head posture in 3D space from 2D image. It is a crucial technique for face recognition, gaze estimation, facial attribute recognition, etc. However, fast head pose estimation executing on the terminal for video edge computation has many challenges due to the computational complexity of the existing algorithms. In this paper, we propose a fast head pose estimation method based on a novel Rotation-Adaptive facial landmark detection powered by Local Binary Feature (RALBF). The landmark detection method is structured through fusing the prior of the rotation information provided by the Progressive Calibration Networks (PCN) face detector to a Local Binary Feature (LBF) based landmark detection method, which improves the robustness against head pose variations and simultaneously keep the computing efficiency. RALBF is trained and tested on 300W dataset and AFLW2000 dataset, it is verified by the accuracy evaluation that RALBF performs better than LBF. To improve the speed of head pose estimation, the 68, 51 and 10 landmarks distribution schemes are explored and compared on speed and accuracy. In the 10 landmarks scheme, the head pose estimation running once only takes 8.3ms on Intel i7-6700HQ CPU and takes 21.8ms on HiSilicon SoC Hi3519AV100, and the average error of Euler angle is 5.9973 • when the face yaw angle is between ±35 • on AFLW2000 3D dataset. Experiments demonstrate our approach performing well on real scenes. INDEX TERMS Head pose estimation, facial landmark detection, PnP problem, local binary features.

show abstract

Section: A Cnn-based Methodsmentioning

confidence: 99%

“…The process can be described as (9) and (11). The affine transformation matrix can be expressed as (8), and the rotation transformation matrix can be expressed as (10).…”

Section: B the Ralbf Landmark Detectionmentioning

confidence: 99%

Fast Head Pose Estimation via Rotation-Adaptive Facial Landmark Detection for Video Edge Computation

et al. 2020

View full text Add to dashboard Cite

show abstract

“…Earlier facial landmark extractors are based on simple machine learning methods such as the ensemble of regression trees (ERT) [22] as in the Dlib package [24]. The more recent ones are based on CNN models, which have achieved significantly improved performance over the traditional methods, e.g., [7,19,45,50,58,65]. The current CNN-based facial landmark extractors typically contain two stages of operations.…”

Section: Facial Landmark Extractorsmentioning

confidence: 99%

Toward the Creation and Obstruction of DeepFakes

Sun

et al. 2022

Handbook of Digital Face Manipulation and Detection

View full text Add to dashboard Cite

AI-synthesized face-swapping videos, commonly known as DeepFakes, is an emerging problem threatening the trustworthiness of online information. The need to develop and evaluate DeepFake detection algorithms calls for large-scale datasets. However, current DeepFake datasets suffer from low visual quality and do not resemble DeepFake videos circulated on the Internet. We present a new large-scale challenging DeepFake video dataset, Celeb-DF, which contains 5, 639 high-quality DeepFake videos of celebrities generated using an improved synthesis process. We conduct a comprehensive evaluation of DeepFake detection methods and datasets to demonstrate the escalated level of challenges posed by Celeb-DF. Then we introduce Landmark Breaker, the first dedicated method to disrupt facial landmark extraction, and apply it to the obstruction of the generation of DeepFake videos. The experiments are conducted on three state-of-the-art facial landmark extractors using our Celeb-DF dataset.

show abstract

“…Penelitian terkait Facial Landmark Detection telah banyak dilakukan oleh beberapa peneliti seperti Xu Zou dkk (2019) mengimplementasikan metode Hierarchical Structured Landmark Ensemble (HSLE) [10]. Pada penelitian lainnya yang dilakukan oleh Jiangjing dkk pada tahun 2017 dijelaskan untuk melakukan deteksi objek pada wajah penulis menggunakan approach hybrid antara regresi dan deep learning yang menghasilkan redudansi data [11].…”

Section: Studi Literatur a State Of The Artunclassified

Implementasi Kombinasi Metode Mean Denoising dan Convolutional Neural Network pada Facial Landmark Detection

Udayana¹,

Supartha²

2021

j. nas. pendidik. teknik. inform.

View full text Add to dashboard Cite

Facial landmark detectionmerupakan bagian dari facial recognition,bertujuan untuk mengidentifikasi titik fokus pada wajah berdasarkan ciri penampakan bagian wajah yang cenderung menonjol, seperti area mata, hidung, bibir, serta tulang pipi. Facial landmark detection sering diimplementasikan pada bidang pengenalan wajah, prediksi pose wajah, rekonstruksi wajah 3 dimensi, serta pengembangan sistem deteksi kelelahan karyawan berdasarkan ekspresi wajah. Seiring bertambahnya ketersediaan citra wajah dan kebutuhan proses komputasi yang cepat, metode Convolutional Neural Network (CNN) diimplementasikan pada facial landmark detection. Namun beragamnya kualitas citra menyebabkan CNN kurang optimal dalam melakukan deteksi. Oleh karena itu guna mengatasi permasalahan terkait kualitas citra ini, diimplementasikan metode mean denoising sebagai upaya peningkatan nilai akurasi CNN dalam melakukan pendeteksian landmark wajah. Dataset citra wajah diperoleh dari platform Kaggle, LFW-People, AFLW200 dan Female Facial Image Dataset, dengan total sebanyak 2.050 citra wajah, dan terbagi menjadi 2.000 data latih dan 50 data uji. Berdasarkan hasil pengujian, kombinasi metode CNN dengan mean denoising menghasilkan peningkatan akurasi yang lebih baik dalam pengenalan objek pada wajah pada kualitas citra yang heterogen dengan rata-rata akurasi pengujian sebesar 81,33%.Akurasi yang cukup baik ini didapatkan karena citra wajah masukan dilakukan penghilangan noise terlebih dahulu sehingga fitur dari citra yang seringkali menyebabkan sistem CNN salah dalam mengidentifikasi objek pada wajah dapat diminimalisir.

show abstract

Learning Robust Facial Landmark Detection via Hierarchical Structured Ensemble

Cited by 57 publications

References 49 publications

Fast Head Pose Estimation via Rotation-Adaptive Facial Landmark Detection for Video Edge Computation

Fast Head Pose Estimation via Rotation-Adaptive Facial Landmark Detection for Video Edge Computation

Toward the Creation and Obstruction of DeepFakes

Implementasi Kombinasi Metode Mean Denoising dan Convolutional Neural Network pada Facial Landmark Detection

Contact Info

Product

Resources

About